Risk and the Gender Pay Gap

Groningen, Non. Jul. MMDCCLXXI A.U.C.,

A former prof of mine recently posted this interesting blog post on the relation between risk and the gender pay gap. In short, he argues the following. As an employer it is easy to not hire women, or to offer lower wage, when they get pregnant, because it is very visible that they are. In other words, there is low uncertainty involved, so it is easy to adjust wages to the situation. Meanwhile, men take more risks on average. With men it is not clear what risk they will take, and what the results of this will be: “The list of such risks is long and stereotypical: lying salesmen, heart-attack stress, over-aggressive purchases,  bad hires based on hunches, sexual harassment, etc. These risks are hard to predict or plan for, so they are more like uncertainties (“known unknowns”) than risks that can be insured against, which means that firms may be “paying too much” for the generic male employee.”

We could also turn this argument upside down and present another hypothesis: men take more risks, and on average are rewarded for this. Meanwhile, women take less risk, and therefore miss out on average on the rewards for taking risk. This does not mean that every man is rewarded for the risk they take. It just means that the size of the rewards outweigh the size of the losses for risk takers.

The potential rewards for taking risks are usually much higher than the potential losses from taking risk. You can never lose more than you own, bankruptcy is always the biggest possible loss. Meanwhile, you can earn multiples of what you own if taking the risk leads to a reward. This is especially the case when it comes to wages. Wages cannot go under €0, right? Meanwhile, it is possible to earn millions of euros. If companies reward risk takers, they cannot punish them more than firing them when the risk takes took the wrong risk. Therefore the average wage of risk takers is probably higher than of more conservative people, even though the risk takers fuck up all the time.

Let’s go through this with a simple example. Assume there are two men and two women, and two types of projects, namely risky and non-risky projects. The risky project has a potential pay-off of €100 and has a probability of failure of 50%. The non-risky project has a potential pay-off of €40, and a probability of failure of 10%. We are assuming no costs here (the company they work for pays for the projects, but the employees can gain part of the pay-off as a bonus, or something). The men launch themselves into the risky project, while the women deliberate and decide on a more conservative course. As it turns out, the women both earn €40, while one man gains €100 and the other loses his job. The average female income is €40, while the average male income is €50. Even though one man lost his job and gains nothing, the average male income is much higher.

While I have not tested this hypothesis, it is not contradicted by income data I could access. Consider the following (fictitious) graph:


We see that on average women earn a lower wage. Some men earn really high wages, while comparatively few women earn wages much higher than the mean average. Whenever there are distributions which are cut-off to one side, and have a long tail to the other side we should expect the mean average (in this case: all income divided by all women or men) to be higher than the median (the person with the income that is in the middle of the distribution: 50% of the people earn more, 50% earn less). The men have a much larger tail to the right than the women in this graph. After all, you cannot earn less than €0 as a wage, while you can earn millions of euros. So the average male income should be further away from the median male income than the female average should be away from the median female income. Moreover, the median male income should be closer to the median female income than the average male income should be to the average female income.

This corresponds to income data for the Netherlands in 2014 and for the UK in 2015-2016. For instance, see the next table for the Netherlands. Median male income is closer to median female income than average male income is to average female income. Meanwhile, average female income is closer to median female income than average male income is to median male income.

Income in Euros in the Netherlands 2014
Average income Median income
Characteristics 1000 euro 1000 euro
Total 32,3 25,7
Men 40,3 33,9
Women 23,9 19,5

The same is true for the UK in 2015-2016:

Income in Pounds in the UK 2015-2016
Median income                   Mean income         
Men 25700 38100
Women 20300 26900

All in all this provides zero evidence that my hypothesis is true, but it seems worthwhile to investigate it at some point (if you do, send the results please?).

Bottom line: Risk could play a role in the gender pay gap. The size of the rewards outweigh the size of the losses for risk takers. If men indeed are risk takers, and women are more conservative, this could lead to a gap in average wages between the sexes.

6 thoughts on “Risk and the Gender Pay Gap

Add yours

  1. Good post on the mean vs median (I agree), but you need to put more emphasis on the principal-agent and bankruptcy problem, i.e., that men get rewarded for success (as does the company) but lose far less if they fail (vs the company). You can use LOTS of data from the financial industry to show their (usually male) employees are paid far too much compared to the risk they bring to their firms, investors and taxpayers.

    Liked by 1 person

  2. Thanks!

    I completely agree, but it’s also phd type research we’re talking here. Spending 3 years on this seems reasonable. It would be very interesting to figure this out though, I’m not sure whether it hasn’t been done already though? Not an expert on the literature, really.


    1. Well, it could be worth a PhD, for sure (look up lit on labor, risk, wages and bankruptcy, obviously); that’s what PhDs do: look for answers to interesting questions 😉


  3. I’m just confused. My understanding of the literature is that once one controls for profession, experience, years working, etc… the pay gap that remains is TINY. Which means whatever risk mechanism we are talking about should be working through career selection, how women get experience relative to men, and their working life cycle rather than through ‘employers paying men more than women do the same job’. That gap, as I understand it is vanishingly small. I know Lucie goes through the data as an exercise in her stats course and it also seems to be the case there.

    Liked by 1 person

      1. @BCZ — the gap is *not* tiny, but depends on marriage status (better than single) and children (worse than childless). Even the cleanest controls (women and men from the same grad program show a gap. You can read the paper here: https://www.aeaweb.org/articles?id=10.1257/jep.31.1.183

        @Joes — my point is that “the process” doesn’t work (uncertainty over risk), even controlling for the jobs m/f work. (There’s a massive gap in where they work, so perhaps — from a social perspective — there’s a good reason to discourage high risk jobs that men now take for wages, but then produce damage, e.g., the financial industry.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Blog at WordPress.com.

Up ↑

%d bloggers like this: